search

EvelynFan / FaceFormer

[CVPR 2022] FaceFormer: Speech-Driven 3D Facial Animation with Transformers
MIT License
778 stars 133 forks source link

size mismatch for VOCASET checkpoint #54

Closed MYTHOFLEGEND closed 1 year ago

MYTHOFLEGEND commented 1 year ago

When I run demo by BIWI checkpoint, it works fine. But when I use VOCASET checkpoint, I get this error:

RuntimeError: Error(s) in loading state_dict for Faceformer:
    size mismatch for obj_vector.weight: copying a param with shape torch.Size([64, 8]) from checkpoint, the shape in current model is torch.Size([64, 6]).

Can somebody help me？Thank you very much！

MYTHOFLEGEND commented 1 year ago

do you know how to solve it？😢 @EvelynFan

MYTHOFLEGEND commented 1 year ago

needs to change count of "--train_subjects" to 8